Orbital Research's Advanced Controls group has a wealth of expertise and design experience across the entire range of control system design philosophies - from highly advanced nonlinear design techniques to nontraditional methods such as artificial neural networks. The Advanced Control Group's real forte, however, is creating innovative solutions to difficult controls engineering problem by combining the different techniques in novel ways. Though each of these techniques can be used in a stand-alone fashion, more powerful and intelligent control systems can be produced when they are designed to operate in concert. These hybrid controllers are capable of effectively controlling systems that are intractable to any single approach and provides a level of flexibility and robustness that cannot be achieved in any other way.
In addition to extensive knowledge of modern state space control design techniques such as optimal and H2/H( control, we possess proprietary, cutting edge nonlinear control algorithms that provide a level of performance, stability and robustness that cannot be matched by any commercially available control algorithm. These proprietary algorithms include computationally efficient adaptive control algorithms that can be implemented in real time for highly dynamic systems and nonlinear control algorithms that greatly expand the number and type of systems that can be controlled.
For systems whose parameters evolve smoothly over their operating range we have several linear adaptive control algorithms including Self Tuning Regulators (STR), Model Reference Adaptive Systems (MRAS) and a computationally efficient Generalized Predictive Control (GPC). Using these algorithms, we have developed control systems for suppression of aircraft wing flutter, missile tracking telescopes and active flow effectors.
For more difficult challenges such as the design of fault tolerant control systems for aircraft, we have a family of nonsmooth control algorithms based upon the principle of feedback domination and Lyapunov stability. These algorithms are not only capable of accommodating systems whose dynamics change in a nonsmooth fashion but can also control unstable nonlinear systems whose Jacobian linearization is uncontrollable and hence cannot be controlled, even locally, by any linear controller. This latter attribute is particularly valuable for the design of controllers for underactuated nonlinear systems such as gun stabilization on a light-weigh, flexible gun mount.
Many systems are not readily amenable to treatment by standard system theoretic approaches. For example, cooperative control of a group of autonomous vehicles lacks the inherent input/output nature assumed by traditional control design approaches. Often, all that can be specified a priori is some desired outcome, not the specific actions of each member of the group. In other cases like distributed control of nonlinear systems, theory is not sufficiently advanced to be useful. In these cases we possess a portfolio of biologically inspired algorithms and artificial intelligence techniques that permit the development of effective, robust and flexible control systems.
One type of biologically inspired algorithm, known as a swarm intelligence-based algorithm, is derived from the observed behaviors of social animals such as ants. This type of algorithm has been proven to be very effective, for instance, in the design of cooperative control algorithms for large groups of unmanned vehicles. In these instances, optimization is not feasible in real time and there is a clear need for the development of decentralized strategies that will enable the vehicles to coordinate effectively. By observing the myriad of ways in which colonies of insects use simple, reactionary behaviors to emerge complex group actions such as the creation of temperature regulated nests or birds flocking, we have extracted simple principles and behaviors which allow the development of group coordination algorithms for applications such as UAV swarm control, cargo handling, data packet routing and data mining in large, federated data bases.
Another type of biologically inspired algorithm is the artificial neural network (ANN). We have developed several ANN reflex control systems whose architecture is based upon the actual neural architecture of a cockroach's escape reflex. The cockroach possesses an incredibly robust escape reflex that has been perfected over millions of years through evolution and can, among other things, successfully evade multiple predators simultaneously and take environmental considerations such as obstacles into account, nearly instantaneously. By mimicking this neural architecture we have developed Autonomous Threat Response (ATR) and collision avoidance systems, targeting algorithms and sensor data fusion algorithms.
Orbital Research's Advanced Control Group offers a full range of control algorithms and control system design services. In addition to our portfolio of control algorithms, we have a suite of proprietary design and analysis tools for the development and customization of control algorithms for specific applications. This suite includes a distributed simulation environment, Hybrid Integrated Virtual Environment (HIVE), which permits the rapid development of high fidelity numerical models and facilitates the interaction between multiple systems including human in the loop (HIL) systems. HIVE interacts with our Advanced Control Toolbox (ACT), which allows the rapid formulation of control laws and algorithms as well as their refinement through optimizing searches such as Genetic Algorithms (GA).